[ViewVC] Diff of: cvs/Coro/Changes

Comparing Coro/Changes (file contents):
Revision 1.516 by root, Wed Jul 22 03:02:07 2009 UTC vs.
Revision 1.621 by root, Mon Nov 18 10:27:53 2013 UTC

…		…
1	Revision history for Perl extension Coro.	1	Revision history for Perl extension Coro.
2		2
3	TODO: should explore PerlIO::coroaio (perl leaks like hell).	3	TODO: should explore PerlIO::coroaio (perl leaks like hell).
4	TODO: maybe implement a default message channel, very much like Erlang's
5	actor model (which is cool in a lot of important aspects (failures!),
6	but very lacking in others (higher level ipc)).
7	TODO: unready_all
8	TODO: myhttpd header parsing
9	TODO: channel->maxsize(newsize)?	4	TODO: channel->maxsize(newsize)?
		5	TODO: __GCC_HAVE_DWARF2_CFI_ASM
		6	TODO: swap_sv, maybe add scope_swap_sv, andallow to unswap by repeat?
		7	TODO: swap_sv not undone in pool
		8
		9	6.33 Mon Nov 18 11:26:27 CET 2013
		10	- do not crash when freeing padlists with holes (in 5.18).
		11	- tentative SVt_BIND 5.19 port/fix.
		12
		13	6.32 Tue Nov 5 15:35:35 CET 2013
		14	- use a new algorithm to derive padlists for perl 5.18. The old
		15	one could lead to 0-pointer accesses inside perl (reported
		16	by Darin McBride).
		17
		18	6.31 Thu May 9 07:39:48 CEST 2013
		19	- Coro::AIO requests would crash if the thread was ready'd
		20	while the request was ongoing.
		21
		22	6.29 Wed May 8 02:55:18 CEST 2013
		23	- when an on_destroy handler destructs the coro currently being
		24	destructed a perl scalar could be accessed after being freed,
		25	likely causing a crash.
		26
		27	6.28 Wed Mar 6 06:58:02 CET 2013
		28	- clean remnants of existing __DIE__ and __WARN__ handlers so
		29	they lose their magic and will not cause segfaults later
		30	(testcase by Andrey Sagulin).
		31	- improved Coro::State documentation a bit.
		32	- Coro::Debug::command now flushes the output.
		33	- add hack detection code for x32 abi, because the braindead slugs
		34	who designed that made it look exactly like x86_64 without
		35	providing proper compile time symbols to test for it. as a result,
		36	this detection cannot work reliably.
		37	- valgrind stack registering was broken.
		38	- do not rely on Time::HiRes anymore in Coro::Debug.
		39
		40	6.23 Fri Dec 7 23:36:37 CET 2012
		41	- use experimental fiber implementation on native windows
		42	perls.
		43	- use sizeof (void *) as multiplication factor for stack sizes,
		44	to accomodate the totally braindamaged microsoft 64 bit "os".
		45	- changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
		46	activeperl 5.16 crashes when PerlIO_define_layer is called due
		47	to some bug in the perl dll, strawberry perl at least passes
		48	the testsuite.
		49	- implement Coro::Handle->peeraddr/host/port, for slightly
		50	improved compatibility with LWP.
		51	- implement 5.17 compatibility by almost blindly applying a
		52	good-looking patch by Father Chrysostomos.
		53	- move stack management functions into libcoro 3.
		54	- libcoro version 3 "released".
		55	- support magic values as timed_io_once args.
		56	- recommend AnyEvent 7+ or EV 4+, also require EV
		57	version 4 or newer for Coro::EV.
		58
		59	6.10 Tue Oct 9 01:14:27 CEST 2012
		60	- updated ecb.h, it had a typo that caused it to not compile on many
		61	big endian systems (reported by many people).
		62	- disable memory fences in ecb.h to improve portability.
		63
		64	6.09 Sat Oct 6 23:25:02 CEST 2012
		65	- Coro::EV I/O watchers were not interruptible by exceptions
		66	(Coro::State::throw) (testcase by sten).
		67	- ->throw now puts threads into the ready queue, as this seems to
		68	be expected by existing code, and code that doesn't cope with spurious
		69	wakeups needs fixing anyway.
		70	- use fd -1 in mmap.
		71	- cast I32 to int in error message printf.
		72	- warn about broken so-called "hardened" kernels.
		73
		74	6.08 Fri Apr 13 12:05:47 CEST 2012
		75	- be more aggressive about exiting like perl does - formerly,
		76	exiting from the non-main thread would not execute END blocks.
		77
		78	6.07 Fri Nov 11 21:21:48 CET 2011
		79	- work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
		80	- update ecb.h.
		81
		82	6.06 Mon Aug 8 23:59:48 CEST 2011
		83	- cygwin unfortunately patches the stack at runtime, so we use the pthreads
		84	backend, which is an order of magnitude slower. unfortunately, cygwins
		85	pthread implementation isn't very complete either, so allocate the stack
		86	twice just to be sure.
		87	(note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
		88	in its ithreaded perl - best recompile cygwin and use the 'w'indows
		89	backend for much better performance. also disable ithreads for
		90	even better performance...).
		91
		92	6.05 Thu Aug 4 21:36:36 CEST 2011
		93	- blush, condvar values would not be propagated from send to recv anymore
		94	(reported by Chip Salzenberg).
		95	- use exponential increase for the readline buffer length in
		96	Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
		97
		98	6.04 Wed Aug 3 17:15:45 CEST 2011
		99	- use even more efficient and more compatible condvars for
		100	compatibility to AnyEvent 6.x :)
		101	- more inconsequential ecb.h updates.
		102
		103	6.03 Wed Aug 3 11:41:30 CEST 2011
		104	- change how Coro patches AnyEvent condvars for compatibility to
		105	AnyEvent 6.x.
		106	- update ecb.h, to no longer include <pthread.h> in case WinNT.h
		107	hasn't been included.
		108
		109	6.02 Wed Jul 13 04:35:19 CEST 2011
		110	- "improve portability to Gentoo" - gentoo manages to put perl variables
		111	in memory areas that are farther than 2gb apart, which the jit couldn't
		112	handle and barfed. now it's just a bit slower on gentoo and similar
		113	systems.
		114
		115	6.01 Sun Jul 3 12:31:14 CEST 2011
		116	- workarounds are good, but the test for whether pthreads are used
		117	was not good. thisone should be better.
		118	- check differently whether gcc generates cfi instructions itself.
		119
		120	6.0 Wed Jun 29 19:43:35 CEST 2011
		121	- INCOMPATIBLE CHANGE: unreferenced coro objects will now be
		122	destroyed and cleaned up automatically (e.g. async { schedule }).
		123	- implement a JIT compiler for part of the thread switch code,
		124	which gives a 50% speed improvement on threaded perls, and
		125	about 4% on non-thraeded perls (so threaded perls now finally
		126	reach about half the speed of non-threaded perls).
		127	- slightly modernise Coro::Intro, add section about rouse functions.
		128	- avoid DEFSV and ERRSV, giving another 10% improvement
		129	in thread switching.
		130	- Coro::State->is_destroyed is now called is_zombie.
		131	- implement a Coro->safe_cancel method that might fail, but
		132	cancels in a "safer" way if it succeeds.
		133	- add preliminary support for DEBUGGING perls.
		134	- get rid of two hash-accesses when initialising a new Coro - this
		135	speeds up coro creation by almost a factor of two.
		136	- croak when a coro that is being cancelled tries to block
		137	(e.g. while executing a guard block), instead of crashing or
		138	deadlocking.
		139	- use a more robust and also faster method to identify Coro::State
		140	objects - speeds up everything a bit.
		141	- implement Coro->cancel in XS for a 20% speed improvement, and to
		142	be able to implement mutual cancellation.
		143	- speed up context switches by a percent or two by more efficiently
		144	allocating context stack entries.
		145	- implement Coro->join and Coro->on_destroy in XS for a speedup and
		146	a reduction in memory use.
		147	- cancelling a coro while it itself is cancelling another coro is
		148	now supported and working, instead of triggering an assertion.
		149	- be a bit more crash-resistant when calling (buggy) on_destroy
		150	callbacks (best effort).
		151	- move on_destroy into the slf_frame, to allow extension slf
		152	functions to have destructors.
		153	- get rid if coro refcounting - simply crash in other interpreter
		154	threads by nulling the pointers on clone.
		155	- simplify warn/die hook handling when loading Coro - the convoluted
		156	logic seems to be no longer neccessary.
		157	- use libecb instead of our own home-grown gcc hacks.
		158	- document alternatives to Coro::LWP. Please use them :)
		159	- work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
		160	sigaltstack. Really. wine suffers from it, erlang suffers from it,
		161	and it's known since at least 2006.
		162
		163	5.372 Wed Feb 23 06:14:30 CET 2011
		164	- apparently mingw doesn't provide a working gettimeofday, try to
		165	work around that by relying on Time::HiRes (indirectly brought to
		166	my attention by Max Maischein).
		167	- fix some portability issues when Time::HiRes was used.
		168
		169	5.371 Mon Feb 21 14:36:08 CET 2011
		170	- backport to windows process emulation code again.
		171
		172	5.37 Sat Feb 19 07:49:44 CET 2011
		173	- add a big "Coro thread life cycle" section to "man Coro".
		174	- try a tentative workaround against the breakage that 5.13 has
		175	introduced without depreciation period. sigh.
		176	- no longer use Time::HiRes if gettimeofday is available, which
		177	saves quite a lot of memory.
		178
		179	5.36 Sun Feb 13 05:33:41 CET 2011
		180	- automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
		181	Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
		182	method call.
		183	- undocument Coro::Timer::sleep and obsolete whole module.
		184	- optimise Coro::Timer::timeout memory and cpu usage.
		185	- slightly updated Coro::Intro for recent changes.
		186	- do not initialise PL_dirty anymore.
		187
		188	5.25 Thu Nov 11 01:08:39 CET 2010
		189	- try a different approach on netbsd - netbsd 5 finally has marginally
		190	working pthreads, but still broken ucontext/sigaltstack.
		191	- openbsd 4.8 finally got their act together, Coro works out of the box
		192	with asm, setjmp and pthreads (no change, just informational).
		193
		194	5.24 Sat Oct 23 11:27:12 CEST 2010
		195	- port to the EV 4.0 API.
		196	- work around bugs in mingw32, making strawberry perl work
		197	out of the box.
		198	- correctly modify Coro::AIO function prototypes
		199	so that they reflect the "no optional parameters" rule.
		200	- "ported" libcoro to C++.
		201
		202	5.23 Mon May 17 18:50:42 CEST 2010
		203	- be more resistant to ordering changes when initialising
		204	Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
		205	Waldorf).
		206	- document that perl 5.12 deliberately removed support for cloning.
		207
		208	5.22 Wed Apr 14 03:55:35 CEST 2010
		209	- correctly return udnef on errors in Coro::Handle::read/write
		210	(testcase by Marc Mims).
		211	- convert Coro::Util into a "perl compatibility wrapper" - the functions
		212	are less useful now, but are drop-in replacements for existing
		213	functions, listing better alternatives in the documentation. This also
		214	fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
		215	with Coro::Util::inet_aton.
		216	- do not override $Coro::idle unconditionally in Coro.pm, as other
		217	modules could have provided their own idle coro already
		218	(for exmaple, Coro::AnyEvent).
		219	- fix Coro::Util::gethost* functions.
		220	- Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
		221
		222	5.21 Wed Dec 16 07:19:51 CET 2009
		223	- automatically load Coro::AnyEvent when AnyEvent and Coro are used
		224	together.
		225	- add some examples on how to combine other event loops with Coro in
		226	Coro::AnyEvent, and how to run it (and not to block). Seems to be
		227	the most common source of confusion.
		228	- try to catch people naively blocking in an event callback.
		229	- work around the perl filehandle bug issue in conjunction with
		230	older common::sense (as indirectly pointed out by ZSystem).
		231	- clarify the "not from signal handlers" section.
		232
		233	5.2 Sun Oct 4 14:54:24 CEST 2009
		234	- Coro::Storable destroyed the prototypes of the functions it wrapped.
		235	- export rouse_cb and rouse_wait by default now.
		236	- fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
		237	- new method $state->swap_sv.
		238	- added section on "windows process emulation" to the manpage,
		239	after a not-so-fruitful (nor-friendly) "discussion" with chip
		240	salzenberg (discussion implies arguments, but his only arguments
		241	were ad-hominems, one wonders why he started it in the first
		242	place). I hope this explains it well enough for him to understand,
		243	and maybe well enough for others to understand.
		244	- use common::sense everywhere now.
		245	- idle callbacks are no longer supported, use idle coros instead.
		246	- print a thread listing when a deadlock is detected.
		247
		248	5.17 Sat Aug 22 23:09:31 CEST 2009
		249	- work around a bug in the perl debugger causing crashes
		250	when running under the debugger by marking _pool_handler as nodebug.
		251	- speed up Coro::async considerably.
		252	- try some hacks to get netbsd to work "more often" - their broken
		253	setjmp/longjmp, ucontext and phtreads are really hard on Coro.
		254	- convert Coro to AE 5.0 API.
		255
		256	5.162 Tue Jul 28 04:04:03 CEST 2009
		257	- perl 5.8.2 is now minimum requirement.
		258	- skip t/19_handle.t on broken windows perls.
10		259
11	5.161 Wed Jul 22 04:47:38 CEST 2009	260	5.161 Wed Jul 22 04:47:38 CEST 2009
12	- Coro::AnyEvent::poll could have a different prototype when EV was	261	- Coro::AnyEvent::poll could have a different prototype when EV was
13	used as backend (analyzed by Tatsuhiko Miyagawa).	262	used as backend (analyzed by Tatsuhiko Miyagawa).
14	- Coro::AnyEvent errornously initialised the event loop when loaded,	263	- Coro::AnyEvent errornously initialised the event loop when loaded,
235	- ->throw is now supported on Coro::State objects.	484	- ->throw is now supported on Coro::State objects.
236	- clean up cctx creation code a bit.	485	- clean up cctx creation code a bit.
237	- entersub is actually a UNOP, not a LOGOP (not a bugfix).	486	- entersub is actually a UNOP, not a LOGOP (not a bugfix).
238		487
239	4.9 Sat Nov 8 17:45:27 CET 2008	488	4.9 Sat Nov 8 17:45:27 CET 2008
240	- (libcoro) did not preserve rbp with CORO_ASM (we are getting there).	489	- (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
241	- (libcoro) no longer leak threads in the experimental pthread backend,	490	- (libcoro) no longer leak threads in the experimental pthread backend,
242	also speed it up considerably.	491	also speed it up considerably.
243	- (libcoro) do not rely on makecontext passing void *'s unscathed.	492	- (libcoro) do not rely on makecontext passing void *'s unscathed.
244	- fix compiletime dependencies on libcoro in the Makefile.	493	- fix compiletime dependencies on libcoro in the Makefile.
245	- cctx_count wasn't always updated properly.	494	- cctx_count wasn't always updated properly.
…		…
307	- remove debugging code related to MgPV_nolen_const, also try to	556	- remove debugging code related to MgPV_nolen_const, also try to
308	make it compile with perl 5.8.6 (yes, apple apparently loves	557	make it compile with perl 5.8.6 (yes, apple apparently loves
309	outdated software). Reported by John S.	558	outdated software). Reported by John S.
310		559
311	4.744 Tue Jul 8 22:06:35 CEST 2008	560	4.744 Tue Jul 8 22:06:35 CEST 2008
312	- correctly provide default DIE/WARN handlers as documented.	561	- correctly provide default DIE/WARN handlers as documented.
313	- also overwrite PL_vtbl_sigelem.svt_clear, even though current	562	- also overwrite PL_vtbl_sigelem.svt_clear, even though current
314	implementations inside perl work fine for us.	563	implementations inside perl work fine for us.
315		564
316	4.743 Mon Jun 16 00:21:57 CEST 2008	565	4.743 Mon Jun 16 00:21:57 CEST 2008
317	- when using Coro::EV without running EV::loop it could	566	- when using Coro::EV without running EV::loop it could
…		…
339	- sprinkle "no warnings" freely over everything, also suppress	588	- sprinkle "no warnings" freely over everything, also suppress
340	warnings for some other modules.	589	warnings for some other modules.
341	- fix typo in WSAEWOULDBLOCK.	590	- fix typo in WSAEWOULDBLOCK.
342		591
343	4.72 Sun May 25 05:14:36 CEST 2008	592	4.72 Sun May 25 05:14:36 CEST 2008
344	- tweak META.yaml a bit, unfortunately, there is no documented way	593	- tweak META.yaml a bit, unfortunately, there is no documented way
345	to have optional dependencies with CPAN. doh :(	594	to have optional dependencies with CPAN. doh :(
346	- avoid running some tests on windows because they would fail due to	595	- avoid running some tests on windows because they would fail due to
347	perl bug (broken fork, broken pipes...).	596	perl bug (broken fork, broken pipes...).
348	- work around perl on windows bugs where perl returns undocumented	597	- work around perl on windows bugs where perl returns undocumented
349	error codes for sysread, syswrite etc. by taking advantage	598	error codes for sysread, syswrite etc. by taking advantage
…		…
353	- use unix domain sockets in testsuite to work around	602	- use unix domain sockets in testsuite to work around
354	common perl implementation bugs on widows (they are emulated by	603	common perl implementation bugs on widows (they are emulated by
355	tcp sockets on windows. ugh.)	604	tcp sockets on windows. ugh.)
356		605
357	4.71 Sat May 24 20:01:27 CEST 2008	606	4.71 Sat May 24 20:01:27 CEST 2008
358	- fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").	607	- fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
359	- take advantage of async name resolution of AnyEvent::Util.	608	- take advantage of async name resolution of AnyEvent::Util.
360	- work around brutal inet_aton override in Coro::LWP.	609	- work around brutal inet_aton override in Coro::LWP.
361	- take advantage of the readyhook in Coro::EV, for smoother	610	- take advantage of the readyhook in Coro::EV, for smoother
362	scheduling.	611	scheduling.
363		612
364	4.7 Sun May 11 00:32:19 CEST 2008	613	4.7 Sun May 11 00:32:19 CEST 2008
365	- completely reworked the Coro manpage.	614	- completely reworked the Coro manpage.
366	- added Coro::AnyEvent, generic event loop integration.	615	- added Coro::AnyEvent, generic event loop integration.
367	- implement cancel, ready and kill commands in Coro::Debug.	616	- implement cancel, ready and kill commands in Coro::Debug.
368	- document find_coro in Coro::Debug.	617	- document find_coro in Coro::Debug.
369	- incompatible change: rename has_stack to has_cctx.	618	- incompatible change: rename has_stack to has_cctx.
370	- Coro::AIO and Coro::BDB no longer force event model detection,	619	- Coro::AIO and Coro::BDB no longer force event model detection,
371	use AnyEvent::AIO and AnyEvent::BDB.	620	use AnyEvent::AIO and AnyEvent::BDB.
…		…
495		744
496	4.11 Thu Oct 11 02:40:24 CEST 2007	745	4.11 Thu Oct 11 02:40:24 CEST 2007
497	- port to threaded perls.	746	- port to threaded perls.
498		747
499	4.1 Thu Oct 11 02:38:16 CEST 2007	748	4.1 Thu Oct 11 02:38:16 CEST 2007
500	- incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now	749	- incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
501	be local to each coro (see Coro::State).	750	be local to each coro (see Coro::State).
502	- incompatible change: for very deep reasons, cede and cede_notself	751	- incompatible change: for very deep reasons, cede and cede_notself
503	cannot return anything, so nothing will be returned.	752	cannot return anything, so nothing will be returned.
504	- possibly bring back 5.10 compatibility (untested).	753	- possibly bring back 5.10 compatibility (untested).
505	- work around stupid (and wrong) warning on 5.10 :(.	754	- work around stupid (and wrong) warning on 5.10 :(.
…		…
577	but stateful semantics.	826	but stateful semantics.
578	- fixed a lot of typos in Coro.pm (patch submitted by David	827	- fixed a lot of typos in Coro.pm (patch submitted by David
579	Steinbrunner, which applied flawlessly).	828	Steinbrunner, which applied flawlessly).
580		829
581	3.6 Sat Apr 14 17:13:31 CEST 2007	830	3.6 Sat Apr 14 17:13:31 CEST 2007
582	- added some bugfixes to get eg/myhttpd working again.	831	- added some bugfixes to get eg/myhttpd working again.
583	- added Coro::Storable for often-cede'ing freeze/thaw.	832	- added Coro::Storable for often-cede'ing freeze/thaw.
584	- try to do a clean exit when a coroutine calls exit	833	- try to do a clean exit when a coroutine calls exit
585	(EXPERIMENTAL).	834	(EXPERIMENTAL).
586	- got rid of indirect call through _coro_init.	835	- got rid of indirect call through _coro_init.
587	- updated the partly antique examples in eg/ to	836	- updated the partly antique examples in eg/ to

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing Coro/Changes (file contents): Revision 1.516 by root, Wed Jul 22 03:02:07 2009 UTC vs. Revision 1.621 by root, Mon Nov 18 10:27:53 2013 UTC

Diff Legend

Comparing Coro/Changes (file contents):
Revision 1.516 by root, Wed Jul 22 03:02:07 2009 UTC vs.
Revision 1.621 by root, Mon Nov 18 10:27:53 2013 UTC